retriever.py

import datasets
from langchain_core.documents import Document

guest_dataset = datasets.load_dataset('agents-course/unit3-invitees', split='train')

docs = [
    Document(
        page_content="\n".join([
            f"Name: {guest['name']}",
            f"Relation: {guest['relation']}",
            f"Description: {guest['description']}",
            f"Email: {guest['email']}"
        ]),
        metadata={'name': guest['name']}
    )
    for guest in guest_dataset
]

from smolagents import Tool
from langchain_community.retrievers import BM25Retriever

class GuestRetrieverTool(Tool):
    name = "guest_info_retriever"
    description = "Retrieves information about gala guests by name or relation."
    inputs = {
        "query": {
            "type": "string",
            "description": "The name or relation of the guest."
        }
    }
    output_type = "string"

    def __init__(self, docs):
        self.initialized = True
        self.retriever = BM25Retriever.from_documents(docs)
        # Optional: limit top-k results
        self.retriever.k = 3

    def forward(self, query: str) -> str:
        # ✅ correct API
        results = self.retriever.get_relevant_documents(query)
        if results:
            return "\n\n".join(doc.page_content for doc in results[:3])
        return "No matching guest info found..."

guest_info_tool = GuestRetrieverTool(docs)

# ⚠️ Don't run agents at import time.
if __name__ == "__main__":
    from smolagents import CodeAgent, InferenceClientModel
    model = InferenceClientModel(model_id="HuggingFaceH4/zephyr-7b-beta")
    alfred = CodeAgent(tools=[guest_info_tool], model=model)
    print(alfred.run("Tell me about guest named 'Lady Ada Lovelace'"))